Systematic simplification methodology applied to an activated sludge reactor model

Abstract

This article proposes a linearization technique for an activated sludge reactor model. Starting with a reduced model using three state variables related to the organic substrate, biomass and oxygen concentrations; the linearization approach is developed and validated. In a first step, sensitivity analysis is used to evaluate the model output sensitivity to changes with respect to the model input variables. The input variables identified as non-influent, can then be discarded, leading to a model simplification. In a second step, locally valid linear model are derived. The linearization is based on approximation of nonlinear terms by linear combinations of variables of interest. In the third step the unknown parameters in the linear terms are identified using an evolutionary algorithm. The main advantage of the proposed strategy is that it conserves the global structure of the original model. The obtained linear model allows easier studies of the state estimation and control schemes problems.